[PlanetCCRMA] living on the (planet)edge...

[Fernando Pablo Lopez-Lezcano]

> > I'm going to be putting more stuff I want to test there shortly,
> > including a newer alsa (the just released 0.9.4), a newer jack and
> > another kernel (based on 2.4.21-rc6).
> 
> Will the newer alsa only be for the 2.4.21 kernel, or can we have them for
> the 4.20-caps kernel too? If not, can you provide .src.rpms?

I'll build for the caps kernel as well and will provide src rpms. It is
not going to be so "on the edge" for now as I have decided to take the
easy way out (for alsa) and not make the alsa devices owned by whoever
is logged in at the moment (why? it is a long story - the behavior will
be the same as it has been so far). 

> > The jack-audio-connection-kit
> > package sets up a tmpfs mount in /etc/fstab so it updates fstab,
> > potentially dangerous... (but the end result is the magical
> > dissapearance, at least in my laptop, of xruns - I can run it down to
> > 64x3 with no problems!).
> 
> I would gladly spend hours driving back and forth to work, making a rescue
> cd, and rebuilding my fstab by hand with nothing but ash and cat, if it
> means I can get my hdsp multiface working and jack kicking ass with it.

I'm not so sure about the hdsp part. Jack should improve its behavior if
you are running /tmp in a journalled filesystem. That was my experience,
at least in my laptop. 

Currently I'm puzzled by the latency behavior on a hyperthreading P4
(new machine I'm building) when booting with the smp kernel. There are
latency spikes that go up to 10/15 msecs when reading/writing to disk.
Those go away if I boot into a up kernel, _or_ if I boot into the smp
kernel with acpi turned off (in that case the kernel only recognizes one
processor). In a real multiprocessor machine with Athlon MP processors
the glitches are not there, same kernel, same alsa...

I guess "imaginary" processors are not as good as real ones :-)
[most probably the ht support in the kernel needs some work]
-- Fernando